On Tue, 30 Jun 1998, Jonathan Stone wrote:

> [telnet BINARY negotiation]
> 
> Ken Hornstein writes:
> 
> >Hmmm .... I dunno.  I mean, if both sides support BINARY, then it's
> >only a win, right?  Or perhaps I don't understand BINARY as well as
> >I thought I did.  The only time that this is a loss that I can see
> >is when people say they can do BINARY, but can't.
> 
> I think we're talking at cross-purposes.
> 
> I think it's a mistake for telnetd to _initiate_ the process which
> turns on BINARY. It shouldn't do that unless the client asks for it,
> or unless the user frobs the tty into a state where BINARY is more
> appropriate.

Can telnet do that?

For reference, the first (and only mention) of binary is a:

td: send do BINARY

in the -D report.

> (If only because it's busted a telnet client which used to work ;).
> 
> If the user asks for binary, sure, let telnetd go ahead.  But in the
> absence of any mechanism for both sides to establish what "BINARY"
> means, telnetd shouldn't be turning on BINARY, just off its own bat.

I think the problem's really that a number of clients, including
MacSamson, will negotiate BINARY but still send NVT codes, which aren't
supposed to be in a BINARY stream. At least as far as we could tell from
the standards.